Skip to content
GFS
~8 min read · 1,858 words ·updated 2026-04-29 · confidence 21%

DCF assumptions — load-bearing inputs

Scope. This file enumerates the inputs that drive a GFS DCF — it does not run a complete valuation model. The triangulated valuation case lives in ../07_thesis/. Use this file as a reference for any per-share fair-value scenario.

Confidence legend: ✓ verified-primary (20-F / 6-K) · ◐ aggregator / market data · ⚠ inferred / estimate / scenario

1. Capital structure / WACC anchor

ComponentValueNotes / Source
Risk-free rate (10Y UST)4.30% ◐April 2026 yield curve; Fed-rate-cycle-adjusted
Equity risk premium5.5% ⚠Mid-cycle US ERP per Damodaran-style framework
Beta (5-yr levered)1.50 ✓Per STOCK_PRICE_DATA.json
Cost of equity = Rf + β × ERP12.55%Computed
Pre-tax cost of debt5.50% ⚠Implied from FY25 interest expense $93M / avg debt ~$1.5B; partially offset by EDB low-rate Singapore loan
After-tax cost of debt (T = 21%)4.35%Computed
Debt / (Debt + Equity) at market values~3.5%Mkt cap $33.1B / Total cap $34.3B
Equity / (Debt + Equity)~96.5%
WACC12.30%Lightly debt-funded; near cost-of-equity

Read on WACC. GFS is structurally near cost-of-equity because of its modest debt load ($1.2B vs $33B mkt cap). Standard analyst convention would use WACC of 11-13% for GFS — at the higher end of semis-comps because of cyclicality, foreign-incorporation tax friction, and Mubadala-controlled discount. Comparison: TSMC WACC is typically 9-10%; UMC 10-11%.

2. Terminal growth assumptions

ScenarioTerminal growth (g)Implied terminal yearComment
Bear2.0%2031+Mature-foundry secular growth = global semis CAGR
Base3.0%2031+Specialty-foundry mix lifts above commodity-foundry average
Bull4.0%2031+AI / Comm-DC mix shift sustains above-average growth

Implied EV/EBITDA exit multiples (using 12.3% WACC):

  • Bear (g=2%): 1 / (12.3% − 2%) = ~9.7x EV/Forward-period-EBITDA
  • Base (g=3%): 1 / (12.3% − 3%) = ~10.7x
  • Bull (g=4%): 1 / (12.3% − 4%) = ~12.1x

These exit multiples are internally consistent with the comp-set trading multiples (UMC ~5.5x, SMIC ~13x, VIS ~8x, Tower ~9.5x, TSMC ~14.5x).

3. Revenue forecast — segment-level scaffolding

Build-up of FY26-FY30 revenue from the FY25 base of $6,791M:

Bear case ⚠

SegmentFY25 ($M)FY26EFY27EFY28EFY29EFY30E5Y CAGR
Smart Mobile2,6782,5002,4002,4002,4502,500−1.4%
Automotive1,4101,5401,6501,7501,8401,920+6.4%
Home/IIoT1,1891,2001,2501,3001,3501,400+3.3%
Comm/Infra/DC7458008809701,0701,180+9.6%
Non-Wafer7698008509009501,000+5.4%
Total6,7916,8407,0307,3207,6608,000+3.3%

Base case (mgmt-aligned trajectory) ⚠

SegmentFY25 ($M)FY26EFY27EFY28EFY29EFY30E5Y CAGR
Smart Mobile2,6782,6502,7002,8002,9003,000+2.3%
Automotive1,4101,6201,8302,0502,2602,470+11.9%
Home/IIoT1,1891,2501,3301,4201,5101,600+6.1%
Comm/Infra/DC7459301,1501,4001,6802,000+21.9%
Non-Wafer7698709701,0701,1801,300+11.1%
Total6,7917,3207,9808,7409,53010,370+8.8%

Bull case ⚠

SegmentFY25 ($M)FY26EFY27EFY28EFY29EFY30E5Y CAGR
Smart Mobile2,6782,7502,9003,0503,2003,400+4.9%
Automotive1,4101,7001,9802,2602,5402,820+14.9%
Home/IIoT1,1891,3001,4201,5401,6701,800+8.6%
Comm/Infra/DC7451,0501,4001,8002,2502,800+30.3%
Non-Wafer7699201,0701,2501,4401,650+16.5%
Total6,7917,7208,7709,90011,10012,470+12.9%

The load-bearing segment is Comm/Infra/DC. In bull case, this segment goes from $745M (FY25) → $2,800M (FY30) at +30% CAGR — driven by hyperscaler co-packaged optics demand, Marvell custom-AI silicon ramps, NLM Photonics + Ayar Labs production scale, and Lightwave Logic electro-optic polymer monetization through 2027-2030. This segment carries materially higher gross margin than the consolidated average.

4. Margin trajectory — the Q4’25 inflection

Per margins and pricing, Q4’25 IFRS GM was 27.8% and non-IFRS GM 29.0%.

YearIFRS GM (analyst)Non-IFRS GM (analyst)Adj. EBITDA margin
FY24 actual24.5%25.3%n/d
FY25 actual24.9%26.1%34.7% (FY25 full)
FY26E base26.5%27.5%35.5%
FY27E base ⚠27.5%28.5%36.5%
FY28E base ⚠28.5%29.5%37.5%
FY29E base ⚠29.0%30.0%38.0%
FY30E base ⚠29.5%30.5%38.5%

Drivers:

  • Mix shift to Auto + Comm/DC + Non-Wafer (mgmt-driven)
  • Sustained cost discipline + opex leverage
  • AMITC depreciation-shield offset starting FY26
  • Photonics-process ASP premium realization

5. Capital intensity (Capex / Sales)

YearCapex ($M)Revenue ($M)Capex / Sales
FY231,804 ✓7,39224.4%
FY24625 ✓6,7509.3% (cycle trough)
FY25722 ✓6,79110.6%
FY26E base1,0007,32013.7%
FY27E base ⚠1,2507,98015.7%
FY28E base ⚠1,4008,74016.0%
FY29E base ⚠1,3009,53013.6%
FY30E base ⚠1,30010,37012.5%

Mid-cycle capex/sales of ~13-16% is consistent with management commentary on capacity buildout for AI / advanced packaging.

Government grant offsets to capex (deferred-income recognition):

  • CHIPS Direct Funding $1.575B over 4-5 years = ~$315-395M annual disbursement (milestone-dependent)
  • NY State $570M over 3-4 years = ~$140-190M annual
  • AMITC tax credit at 35% on AMITC-eligible capex (likely ~70% of total capex) = effective tax credit ~$200-300M annual on $1.0-1.4B capex
  • Total annual subsidy / credit: $650-900M vs gross capex $1.0-1.4B → net capex of $200-500M annually (very capital-light effective)

The CHIPS / AMITC government-subsidy stack is the single most important “cash” input to a GFS DCF. It transforms the DCF from a “capital-intensive specialty foundry” model to something approaching “asset-light specialty supplier” — with effective net capex of just 4-7% of sales after subsidies.

6. Tax rate

GFS is Cayman-incorporated (Cayman → no corporate tax) but generates income across US, Germany, Singapore, and other jurisdictions. Effective tax rate FY25: 2.5% ($23M tax / $911M pre-tax income) — extremely low due to:

  • AMITC tax credit refund timing
  • Singapore tax incentives (EDB-supported)
  • Cayman Islands tax neutrality at parent level
  • US tax loss carryforwards from FY24

Forward tax rate assumptions:

  • FY26E: ~5-8% ⚠ (AMITC enhanced; some normalization)
  • FY27E: ~8-12% ⚠ (CHIPS / AMITC steady-state)
  • FY28E+: ~12-18% ⚠ (more normal effective rate)

OBBBA (signed July 2025) provides additional tax shielding via:

  • 100% accelerated depreciation for qualifying property placed in service 2025+
  • Immediate R&D expense deduction (vs prior 5-year amortization)
  • §163(j) interest deduction modifications

Read. GFS’s effective tax rate is structurally low (2-12% range) for the forecast period, providing meaningful EPS uplift. Watch for OBBBA-related disclosures in 2026 6-Ks.

7. Working capital and cash conversion

  • Days sales outstanding (DSO): est. ~50-60 days based on FY25 receivables
  • Days inventory outstanding (DIO): est. ~80-100 days (foundry typical)
  • Days payable outstanding (DPO): est. ~70-85 days
  • Cash conversion cycle: est. ~60-75 days
  • Working capital / sales: ~15-18% sustained

These metrics are typical for a 200/300mm specialty foundry; they suggest working capital should grow approximately in proportion to revenue with no major deleveraging or buildup events expected.

8. Per-share fair-value range — illustrative scenarios

Illustrative only — full triangulation lives in valuation framework (file not yet written).

ScenarioFY28E revenueFY28E adj. EBITDAExit EV/EBITDAImplied FY28 EV ($B)Implied mkt cap ($B)Per share ($)vs. $59.49 spot
Bear7.322.69x23.426.2$46−23%
Base8.743.311x36.339.1$70+18%
Bull9.903.913x50.753.5$96+61%

Probability-weighted (20% Bear / 50% Base / 30% Bull): 0.20 × $46 + 0.50 × $70 + 0.30 × $96 = $72.90/share ⚠ — implying ~22% upside vs $59.49 spot.

Read. Probability-weighted valuation is moderately above spot, anchored on the AI/photonics-driven Comm/Infra/DC segment doubling+ over 5 years. The bear case (−23%) is real and reflects (a) Smart Mobile compression accelerates, (b) photonics ramp slower-than-modeled, (c) ICFR remediation adversarial. The bull case (+61%) requires Comm/Infra/DC to reach $1.8B+ FY28 — which assumes Marvell + Ayar Labs + LWLG-enabled customers all hit full-volume production at GFS by 2028.

9. Open items / backfill queue

  1. Sell-side DCF reconstruction — extract analyst notes (TD Cowen, Citi, JPM, Goldman, BofA, MS, SIG) and reconcile WACC + terminal growth assumptions across the desk.
  2. Beta calculation primary-sourceSTOCK_PRICE_DATA.json reports beta 1.50; reconcile vs Bloomberg / Yahoo / Refinitiv betas.
  3. Per-segment gross margin assumptions — currently inferred at corporate level; sub-segment GM extraction from Industry Analyst Day or sell-side disaggregations.
  4. Capex by fab / by program — not disclosed publicly; private channel checks would refine.
  5. OBBBA quantification — the FY25 20-F notes the policy change but does not disclose dollar quantum of expected benefit. Would refine forward tax-rate forecast.

Sources

  • FY2025 20-F, acc. 0001709048-26-000022, filed 2026-02-27 — historical financials, Note 3 income tax disclosures, OBBBA reference.
  • Q4’25 6-K, acc. 0001709048-26-000012, filed 2026-02-11 — FY25 full-year actuals, Q1 2026 guidance.
  • STOCK_PRICE_DATA.json — spot, beta.
  • CBOE-derived options data (STOCK_OPTIONS_DATA.json, refreshed 2026-04-29) — option-implied vol input to DCF cross-check.

Cross-references